Observations from multisensory body illusions indicate that the body representation can be adapted to changing task demands, e.g., it can be expanded to integrate external objects based on current sensorimotor experience (embodiment). While the mechanisms that promote embodiment have been studied extensively in earlier work, the opposite phenomenon of, removing an embodied entity from the body representation (i.e., disembodiment) has received little attention yet. The current study addressed this phenomenon and drew inspiration from the partial reinforcement extinction effect in instrumental learning which suggests that behavior is more resistant to extinction when reinforcement is delivered irregularly. In analogy to this, we investigated whether experiencing occasional visuo-motor mismatches during the induction phase of the moving rubber hand illusion (intermittent condition) would result in slower disembodiment as compared to a regular induction phase where motor and visual signals always match (continuous condition). However, we did not find an effect of reinforcement schedule on disembodiment. Keeping a recently embodied entity in the body schema, therefore, requires constant updating through correlated perceptual and motor signals.

Awareness of social biases is crucial as they impact both individual behavior and societal outcomes. Whereas previous research indicates that self-awareness of ego-nonthreatening biases enhances self-regulation, the effects of self-awareness of ego-threatening biases remain underexplored. Preliminary findings suggest that awareness of ego-threatening biases related to rumination may lead to maladaptive states. However, these findings await replication with standardized instruments. To address this gap, we conducted two studies. In Study 1 (N = 1609), we developed and validated the 12-item Self-Awareness of Ego-Threatening Biases Questionnaire (SAETBQ). Consistent with our hypotheses, self-awareness of ego-threatening biases (as measured by the SAETBQ) correlated with higher moral disengagement, lower self-diagnostic motive, and lower integrative self-knowledge, indicating a tendency towards ego deterioration, whereas self-awareness of ego-nonthreatening biases (as measured by the Metacognitive Self questionnaire) showed the opposite pattern of correlations, indicating a tendency towards beneficial self-regulation. In Study 2 (N = 681), Dark Triad traits correlated positively and Light Triad traits negatively with self-awareness of ego-threatening biases. These results underscore the complex role of self-awareness in managing cognitive biases.

The maximization of submodular functions under various kinds of constraints is a central component of many combinatorial optimization problems, since submodular functions naturally cover the property of diminishing returns. This dissertation comprises four scientific articles in which we consider three different combinatorial optimization problems involving the maximization of submodular functions under knapsack or cardinality constraints. The first article considers the maximization of a submodular function defined on a weighted set of items under a knapsack constraint with unknown capacity. Assume that items are packed sequentially into the knapsack, and that an oracle reveals whether an item being attempted for packing fits into the currently packed knapsack. If an item fits, it is packed irrevocably; otherwise, either packing stops immediately (packing without discarding), or the item is removed, and packing continues (packing with discarding). Our main result concerns non-adaptive packing without discarding, under the assumption that the unknown knapsack capacity is greater than or equal to the weight of the heaviest item. Specifically, we present the first polynomial-time algorithm for computing a universal policy that, for any unknown capacity, performs at least as well as the classical greedy algorithm, studied by Wolsey (1982), for the same known capacity. In the second article, we study a game-theoretic variant of maximizing a submodular function under a cardinality constraint. In this variant, an initial solution to the classical problem is determined first. Subsequently, a predetermined number of elements of the ground set, possibly containing elements of the initial solution, are deleted. If any deleted elements were part of the initial solution, they are replaced by a set of at most equal cardinality. The objective is to maximize the value of the ultimate solution, with the deletion being maximally disadvantageous to it. We analyze several special cases of this problem and present polynomial-time algorithms for computing optimal or approximately optimal ultimate solutions. For the general case, we present a polynomial-time algorithm whose approximation guarantee depends on the curvature of the submodular objective function. The last two articles of this dissertation address the classic problem of maximizing a submodular function under a knapsack constraint. The first of these articles focuses on exact solvers: We present a branch-and-bound algorithm along with several acceleration techniques. We compare it against two solvers by Sakaue and Ishihata (2018), which currently achieve the strongest performance reported in the literature, as well as a branch-and-cut algorithm implemented using Gurobi that solves a binary linear reformulation of the submodular knapsack problem, demonstrating that our methods are highly successful. The last article considers variants of the classical greedy algorithm for submodular maximization under a knapsack constraint studied by Wolsey (1982). While the classical algorithm assumes access to an exact incremental oracle in every iteration, we generalize the known approximation results for this algorithm to the presence of only an $\alpha$-approximate oracle that returns in every iteration an item whose relative marginal gain approximates the maximum relative marginal gain by at least $\frac{1}{\alpha}$, with $\alpha \geq 1$ fixed. We also present an approximation result for a variant of the classical greedy algorithm that uses an approximate oracle only in the first iteration and an exact oracle thereafter.

Norming of psychological tests is crucial for the accurate interpretation of test scores. Conventional norming, which relies on subgroups, may introduce bias and requires large samples (uneconomic) to achieve high precision (i.e., low standard errors) of the estimated norm scores. Continuous norming has been proposed as a solution to reduce bias and resolve the dilemma between economy and precision. Continuous norming estimates norm scores based on the entire normative samples – rather than subgroups – using (non-linear) regression. The aim of this dissertation is to examine continuous norming methods to improve both test development and psychological diagnostics. To this end, this dissertation comprises four research studies. This first study includes both a systematic review of continuous norming and an empirical study. The systematic review introduces different continuous norming methods and highlights their respective advantages and limitations. The empirical study compares the precision of conventional and continuous norms and investigates the presentation of continuous norms in classical norm tables. The second study provides a systematic overview of German-language tests, showing that continuous norming methods are rarely applied and that descriptions of the norming process in test manuals are often scarce. Such a scarce reporting hinders a critical evaluation of the estimated norm scores. To address this issue, the third study introduces guidelines for reporting on norm-referenced scores. The guidelines cover relevant aspects along the entire norming process and provide detailed information on each aspect. One aspect, where the guidance is limited, is determining the required sample size. The fourth study addresses this gap by providing empirical based guidance to determine the required sample size for continuous norming. For a comprehensive evaluation of continuous norming, I combined the findings of these studies with (1) an illustration of the effects of bias and precision on individual diagnostics and the financial costs of normative studies, (2) an updated systematic literature search, and (3) an examination of applied norming practices in recently published tests. This integration allowed the extraction of seven best practices for continuous norming. The first best practice is to favor continuous over conventional norming, as it effectively resolves the economy-precision dilemma: continuous norming produces less biased norms than conventional norming and requires smaller sample sizes to achieve high precision. Despite these advantages, continuous norming is still rarely applied in German-language tests. The guidelines and sample-size recommendations provided in this dissertation may facilitate a wider adoption of continuous norming methods, thereby improving both test development and psychological diagnostics.

Expectations play a central role in financial markets, yet investors often disagree about the economy’s future. Such disagreement has long been regarded as a potential driver of asset prices, but it remains uncertain whether it reflects mispricing or a priced source of risk. This study addresses the issue by constructing monthly disagreement indices from Consensus Economics forecasts across 24 OECD markets. Firm-level exposure to economic disagreement is estimated using return regressions. The results reveal pronounced cross-country heterogeneity. In developed markets, particularly the United States, greater exposure to disagreement consistently predicts lower future returns, supporting the mispricing hypothesis. In smaller markets, the evidence is mixed, with some cases indicating positive risk premia and others showing no significant effect. Overall, the findings provide new international evidence that the pricing of forecast disagreement is context-dependent, shaped by market structure and institutional depth.